Working Group I: The Scientific Basis

Other reports in this collection Thermohaline circulation changes

Figure 9.21: Simulated water-volume transport change of the Atlantic “conveyor belt” (Atlantic overturning) in a range of global warming scenarios computed by different climate research centres. Shown is the annual mean relative to the mean of the years (1961 to 1990) (Unit: SV, 106 m3s-1). The past forcings are only due to greenhouse gases and aerosols. The future-forcing scenario is the IS92a scenario. See Table 9.1 for more information on the individual models used here.

In the SAR, it was noted that the thermohaline circulation (THC) weakens as CO2 increases in the atmosphere in most coupled climate model integrations. The weakening of the THC is found in both the Northern and Southern Hemispheres. The amount of weakening varied from model to model, but in some cases it was noted that the THC in the North Atlantic stopped completely (Manabe and Stouffer, 1994; Hirst 1999). The weakening of the THC in the Atlantic Ocean results in a reduction of the poleward heat transport that in turn leads to a minimum in the surface warming in the northern North Atlantic Ocean and/or in the circumpolar Ocean (see Section 9.3.2). The reduction in the warming in the North Atlantic region touches the extreme north-eastern part of North America and north-west Europe. The shutting off of the THC in either hemisphere could have long-term implications for climate. However, even in models where the THC weakens, there is still a warming over Europe. For example, in all AOGCM integrations where the radiative forcing is increasing, the sign of the temperature change over north-west Europe is positive (see Figure 9.10).

Figure 9.21 shows a comparison of the strength of the THC through a number of transient experiments with various models and warming scenarios over the 21st century. The initial (control state) absolute strength of the Atlantic thermohaline circulation (THC) varies by more than a factor of 2 between the models, ranging from 10 to 30 Sv (1 Sv = 106 m3s-1). The cause of this wide variation is unclear, but it must involve the sub-grid parametrization schemes used for mixing in the oceans (Bryan, 1987) and differences in the changes of the surface fluxes. The sensitivity of the THC to changes in the radiative forcing is also quite different between the models. Generally as the radiative forcing increases, most models show a reduction of THC. However, some models show only a small weakening of the THC and one model (ECHAM4/OPYC; Latif et al., 2000) has no weakening in response to increasing greenhouse gases, as does the NCAR CSM as documented by Gent (2001). The exact reasons for the difference in the THC responses are unknown, but the role of the surface fluxes is certainly part of the reasons for the differences in the response (see below).

Stocker and Schmittner (1997), using an intermediate complexity model, found that the North Atlantic THC shut-down when the rate of 1%/yr of CO2 increase was held fixed for approximately 100 years. This is in agreement with the earlier AOGCM study of Manabe and Stouffer (1994), where the THC shut-down in an integration where the CO2 concentration increased by 1%/yr to four times its initial value. In integrations where the CO2 stabilised at doubling, the THC did not shut-down in either study (Stocker and Schmittner 1997; Manabe and Stouffer 1994). Furthermore, in the Manabe and Stouffer (1994) AOGCM where the CO2 is stabilised at four times its normal value, the THC recovers to the control integration value around model year 2300. A recent study (Stouffer and Manabe, 1999) found that the amount of weakening of the THC by the time of CO2 doubling is a function of the rate of CO2 increase and not the absolute increase in the radiative forcing. They found the slower the rate of increase, the more the weakening of the THC by the time of CO2 doubling.

Figure 9.22: Time-series of the zonally integrated Atlantic mass transport stream function at 30°N and 1500 m depth, close to the maximum of the stream function simulated by the (a) ECHAM3/LSG model and the (b) GFDL_R15_b model. For a description of the experiments see Table 9.5.

The evolution of the THC in response to future forcing scenarios is a topic requiring further study. It should be noted in particular that these climate model experiments do not currently include the possible effects of significant freshwater input arising from changes in land ice sheets (Greenland and Antarctic ice caps) and mountain glaciers, which might well lead to bigger reductions in the THC. It is too early to say with confidence whether irreversible shut-down of the THC is likely or not, or at what threshold it might occur. Though no AOGCM to date has shown a shut-down of the THC by the year 2100, climate changes over that period may increase the likelihood during subsequent centuries, though this is scenario-dependent. The realism of the representation of oceanic mechanisms involved in the THC changes also needs to be carefully evaluated in the models.

Role of the surface fluxes
The role of heat, fresh water and momentum fluxes in weakening the North Atlantic THC as a consequence of increasing atmospheric CO2 concentration has been studied in two different AOGCMs (ECHAM3/LSG, Mikolajewicz and Voss, 2000; and GFDL_R15_b, Dixon et al., 1999). In both these studies (Figure 9.22), two baseline integrations are performed; a control integration in which the CO2 is held fixed, and a perturbation integration in which the CO2 is increasing. The water fluxes from both of these integrations are archived and used as input in two new integrations.

In the first integration, the atmospheric CO2 concentration is held fixed and the fresh water fluxes into the ocean are prescribed as those obtained from the perturbation integration. In the second integration, the CO2 increases as in the perturbation integration and the water fluxes are prescribed to be the fluxes from the control integration (see Table 9.5). In this way, the relative roles of the fresh water and heat fluxes can be evaluated (Figure 9.22).

Table 9.5: The THC-sensitivity experiments.
Experiment CO2 concentration Freshwater flux Wind stress
FSS fixed present day simulated simulated
ISS increasing simulated simulated
IFS increasing from FSS simulated
FIS fixed present day from ISS simulated
FSI fixed present day simulated from ISS
IFF increasing from FSS from ISS

Other reports in this collection

IPCC Homepage